Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Thu, Nov 09, 2017 at 09:32:14AM +0000, Sebastian Blajszczak wrote:
> Hello Roland,
>
> yes both are diskless. But everything is ok and the LVS still exists. /etc/drbd.d/drbdctrl.res is correct :
>
> resource .drbdctrl {
> net {
> cram-hmac-alg sha256;
> shared-secret "5M3oyND1XHtRkKCR5Jdc";
> allow-two-primaries no;
> }
> volume 0 {
> device minor 0;
> disk /dev/drbdpool/.drbdctrl_0;
> meta-disk internal;
> }
> volume 1 {
> device minor 1;
> disk /dev/drbdpool/.drbdctrl_1;
> meta-disk internal;
> }
> on luke {
> node-id 0;
> address 192.168.11.1:6999;
> }
> on leia {
> node-id 1;
> address 192.168.11.2:6999;
> }
> connection-mesh {
> hosts luke leia;
> net {
> protocol C;
> }
> }
> }
>
> So I think I have only to init with drbdmanage the server? Or there something more to do?
This would remove the control volume and recreate it. Assuming you
answered "yes" often enough and the current state, you most likely end
up with loosing your cluster database. Not such a good idea. If you then
start the whole - now fresh/empty - thing, it even removes your res
files. Also not what you want.
> When the sync is OK again, I will update the second server from
> proxmox 4.4 to 5.1 and get the same problems again when I have to
> initilize the drbdmanage.
Get the control volumes fixed *now*. So you say the LVMs are there and
still they are diskless. "drbdadm adjust/attach .drbdctrl". What does
that tell you? If in a two node setup both nodes are diskless, something
is very wrong.
As I probably already wrote: Forget about the other stuff, get DRBD on
the control volume in a sane state.
Regards, rck